| Current Path : /sys/amd64/compile/hs32/modules/usr/src/sys/modules/drm/radeon/@/ofed/drivers/infiniband/ulp/sdp/ |
FreeBSD hs32.drive.ne.jp 9.1-RELEASE FreeBSD 9.1-RELEASE #1: Wed Jan 14 12:18:08 JST 2015 root@hs32.drive.ne.jp:/sys/amd64/compile/hs32 amd64 |
| Current File : //sys/amd64/compile/hs32/modules/usr/src/sys/modules/drm/radeon/@/ofed/drivers/infiniband/ulp/sdp/sdp_cma.c |
/*
* Copyright (c) 2006 Mellanox Technologies Ltd. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
* General Public License (GPL) Version 2, available from the file
* COPYING in the main directory of this source tree, or the
* OpenIB.org BSD license below:
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above
* copyright notice, this list of conditions and the following
* disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials
* provided with the distribution.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* $Id$
*/
#include "sdp.h"
#define SDP_MAJV_MINV 0x22
SDP_MODPARAM_SINT(sdp_link_layer_ib_only, 1, "Support only link layer of "
"type Infiniband");
enum {
SDP_HH_SIZE = 76,
SDP_HAH_SIZE = 180,
};
static void
sdp_qp_event_handler(struct ib_event *event, void *data)
{
}
static int
sdp_get_max_dev_sge(struct ib_device *dev)
{
struct ib_device_attr attr;
static int max_sges = -1;
if (max_sges > 0)
goto out;
ib_query_device(dev, &attr);
max_sges = attr.max_sge;
out:
return max_sges;
}
static int
sdp_init_qp(struct socket *sk, struct rdma_cm_id *id)
{
struct ib_qp_init_attr qp_init_attr = {
.event_handler = sdp_qp_event_handler,
.cap.max_send_wr = SDP_TX_SIZE,
.cap.max_recv_wr = SDP_RX_SIZE,
.sq_sig_type = IB_SIGNAL_REQ_WR,
.qp_type = IB_QPT_RC,
};
struct ib_device *device = id->device;
struct sdp_sock *ssk;
int rc;
sdp_dbg(sk, "%s\n", __func__);
ssk = sdp_sk(sk);
ssk->max_sge = sdp_get_max_dev_sge(device);
sdp_dbg(sk, "Max sges: %d\n", ssk->max_sge);
qp_init_attr.cap.max_send_sge = MIN(ssk->max_sge, SDP_MAX_SEND_SGES);
sdp_dbg(sk, "Setting max send sge to: %d\n",
qp_init_attr.cap.max_send_sge);
qp_init_attr.cap.max_recv_sge = MIN(ssk->max_sge, SDP_MAX_RECV_SGES);
sdp_dbg(sk, "Setting max recv sge to: %d\n",
qp_init_attr.cap.max_recv_sge);
ssk->sdp_dev = ib_get_client_data(device, &sdp_client);
if (!ssk->sdp_dev) {
sdp_warn(sk, "SDP not available on device %s\n", device->name);
rc = -ENODEV;
goto err_rx;
}
rc = sdp_rx_ring_create(ssk, device);
if (rc)
goto err_rx;
rc = sdp_tx_ring_create(ssk, device);
if (rc)
goto err_tx;
qp_init_attr.recv_cq = ssk->rx_ring.cq;
qp_init_attr.send_cq = ssk->tx_ring.cq;
rc = rdma_create_qp(id, ssk->sdp_dev->pd, &qp_init_attr);
if (rc) {
sdp_warn(sk, "Unable to create QP: %d.\n", rc);
goto err_qp;
}
ssk->qp = id->qp;
ssk->ib_device = device;
ssk->qp_active = 1;
ssk->context.device = device;
sdp_dbg(sk, "%s done\n", __func__);
return 0;
err_qp:
sdp_tx_ring_destroy(ssk);
err_tx:
sdp_rx_ring_destroy(ssk);
err_rx:
return rc;
}
static int
sdp_connect_handler(struct socket *sk, struct rdma_cm_id *id,
struct rdma_cm_event *event)
{
struct sockaddr_in *src_addr;
struct sockaddr_in *dst_addr;
struct socket *child;
const struct sdp_hh *h;
struct sdp_sock *ssk;
int rc;
sdp_dbg(sk, "%s %p -> %p\n", __func__, sdp_sk(sk)->id, id);
h = event->param.conn.private_data;
SDP_DUMP_PACKET(sk, "RX", NULL, &h->bsdh);
if (!h->max_adverts)
return -EINVAL;
child = sonewconn(sk, SS_ISCONNECTED);
if (!child)
return -ENOMEM;
ssk = sdp_sk(child);
rc = sdp_init_qp(child, id);
if (rc)
return rc;
SDP_WLOCK(ssk);
id->context = ssk;
ssk->id = id;
ssk->socket = child;
ssk->cred = crhold(child->so_cred);
dst_addr = (struct sockaddr_in *)&id->route.addr.dst_addr;
src_addr = (struct sockaddr_in *)&id->route.addr.src_addr;
ssk->fport = dst_addr->sin_port;
ssk->faddr = dst_addr->sin_addr.s_addr;
ssk->lport = src_addr->sin_port;
ssk->max_bufs = ntohs(h->bsdh.bufs);
atomic_set(&ssk->tx_ring.credits, ssk->max_bufs);
ssk->min_bufs = tx_credits(ssk) / 4;
ssk->xmit_size_goal = ntohl(h->localrcvsz) - sizeof(struct sdp_bsdh);
sdp_init_buffers(ssk, rcvbuf_initial_size);
ssk->state = TCPS_SYN_RECEIVED;
SDP_WUNLOCK(ssk);
return 0;
}
static int
sdp_response_handler(struct socket *sk, struct rdma_cm_id *id,
struct rdma_cm_event *event)
{
const struct sdp_hah *h;
struct sockaddr_in *dst_addr;
struct sdp_sock *ssk;
sdp_dbg(sk, "%s\n", __func__);
ssk = sdp_sk(sk);
SDP_WLOCK(ssk);
ssk->state = TCPS_ESTABLISHED;
sdp_set_default_moderation(ssk);
if (ssk->flags & SDP_DROPPED) {
SDP_WUNLOCK(ssk);
return 0;
}
if (sk->so_options & SO_KEEPALIVE)
sdp_start_keepalive_timer(sk);
h = event->param.conn.private_data;
SDP_DUMP_PACKET(sk, "RX", NULL, &h->bsdh);
ssk->max_bufs = ntohs(h->bsdh.bufs);
atomic_set(&ssk->tx_ring.credits, ssk->max_bufs);
ssk->min_bufs = tx_credits(ssk) / 4;
ssk->xmit_size_goal =
ntohl(h->actrcvsz) - sizeof(struct sdp_bsdh);
ssk->poll_cq = 1;
dst_addr = (struct sockaddr_in *)&id->route.addr.dst_addr;
ssk->fport = dst_addr->sin_port;
ssk->faddr = dst_addr->sin_addr.s_addr;
soisconnected(sk);
SDP_WUNLOCK(ssk);
return 0;
}
static int
sdp_connected_handler(struct socket *sk, struct rdma_cm_event *event)
{
struct sdp_sock *ssk;
sdp_dbg(sk, "%s\n", __func__);
ssk = sdp_sk(sk);
SDP_WLOCK(ssk);
ssk->state = TCPS_ESTABLISHED;
sdp_set_default_moderation(ssk);
if (sk->so_options & SO_KEEPALIVE)
sdp_start_keepalive_timer(sk);
if ((ssk->flags & SDP_DROPPED) == 0)
soisconnected(sk);
SDP_WUNLOCK(ssk);
return 0;
}
static int
sdp_disconnected_handler(struct socket *sk)
{
struct sdp_sock *ssk;
ssk = sdp_sk(sk);
sdp_dbg(sk, "%s\n", __func__);
SDP_WLOCK_ASSERT(ssk);
if (sdp_sk(sk)->state == TCPS_SYN_RECEIVED) {
sdp_connected_handler(sk, NULL);
if (rcv_nxt(ssk))
return 0;
}
return -ECONNRESET;
}
int
sdp_cma_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
{
struct rdma_conn_param conn_param;
struct socket *sk;
struct sdp_sock *ssk;
struct sdp_hah hah;
struct sdp_hh hh;
int rc = 0;
ssk = id->context;
sk = NULL;
if (ssk)
sk = ssk->socket;
if (!ssk || !sk || !ssk->id) {
sdp_dbg(sk,
"cm_id is being torn down, event %d, ssk %p, sk %p, id %p\n",
event->event, ssk, sk, id);
return event->event == RDMA_CM_EVENT_CONNECT_REQUEST ?
-EINVAL : 0;
}
sdp_dbg(sk, "%s event %d id %p\n", __func__, event->event, id);
switch (event->event) {
case RDMA_CM_EVENT_ADDR_RESOLVED:
sdp_dbg(sk, "RDMA_CM_EVENT_ADDR_RESOLVED\n");
if (sdp_link_layer_ib_only &&
rdma_node_get_transport(id->device->node_type) ==
RDMA_TRANSPORT_IB &&
rdma_port_get_link_layer(id->device, id->port_num) !=
IB_LINK_LAYER_INFINIBAND) {
sdp_dbg(sk, "Link layer is: %d. Only IB link layer "
"is allowed\n",
rdma_port_get_link_layer(id->device, id->port_num));
rc = -ENETUNREACH;
break;
}
rc = rdma_resolve_route(id, SDP_ROUTE_TIMEOUT);
break;
case RDMA_CM_EVENT_ADDR_ERROR:
sdp_dbg(sk, "RDMA_CM_EVENT_ADDR_ERROR\n");
rc = -ENETUNREACH;
break;
case RDMA_CM_EVENT_ROUTE_RESOLVED:
sdp_dbg(sk, "RDMA_CM_EVENT_ROUTE_RESOLVED : %p\n", id);
rc = sdp_init_qp(sk, id);
if (rc)
break;
atomic_set(&sdp_sk(sk)->remote_credits,
rx_ring_posted(sdp_sk(sk)));
memset(&hh, 0, sizeof hh);
hh.bsdh.mid = SDP_MID_HELLO;
hh.bsdh.len = htonl(sizeof(struct sdp_hh));
hh.max_adverts = 1;
hh.ipv_cap = 0x40;
hh.majv_minv = SDP_MAJV_MINV;
sdp_init_buffers(sdp_sk(sk), rcvbuf_initial_size);
hh.bsdh.bufs = htons(rx_ring_posted(sdp_sk(sk)));
hh.localrcvsz = hh.desremrcvsz = htonl(sdp_sk(sk)->recv_bytes);
hh.max_adverts = 0x1;
sdp_sk(sk)->laddr =
((struct sockaddr_in *)&id->route.addr.src_addr)->sin_addr.s_addr;
memset(&conn_param, 0, sizeof conn_param);
conn_param.private_data_len = sizeof hh;
conn_param.private_data = &hh;
conn_param.responder_resources = 4 /* TODO */;
conn_param.initiator_depth = 4 /* TODO */;
conn_param.retry_count = SDP_RETRY_COUNT;
SDP_DUMP_PACKET(NULL, "TX", NULL, &hh.bsdh);
rc = rdma_connect(id, &conn_param);
break;
case RDMA_CM_EVENT_ROUTE_ERROR:
sdp_dbg(sk, "RDMA_CM_EVENT_ROUTE_ERROR : %p\n", id);
rc = -ETIMEDOUT;
break;
case RDMA_CM_EVENT_CONNECT_REQUEST:
sdp_dbg(sk, "RDMA_CM_EVENT_CONNECT_REQUEST\n");
rc = sdp_connect_handler(sk, id, event);
if (rc) {
sdp_dbg(sk, "Destroying qp\n");
rdma_reject(id, NULL, 0);
break;
}
ssk = id->context;
atomic_set(&ssk->remote_credits, rx_ring_posted(ssk));
memset(&hah, 0, sizeof hah);
hah.bsdh.mid = SDP_MID_HELLO_ACK;
hah.bsdh.bufs = htons(rx_ring_posted(ssk));
hah.bsdh.len = htonl(sizeof(struct sdp_hah));
hah.majv_minv = SDP_MAJV_MINV;
hah.ext_max_adverts = 1; /* Doesn't seem to be mandated by spec,
but just in case */
hah.actrcvsz = htonl(ssk->recv_bytes);
memset(&conn_param, 0, sizeof conn_param);
conn_param.private_data_len = sizeof hah;
conn_param.private_data = &hah;
conn_param.responder_resources = 4 /* TODO */;
conn_param.initiator_depth = 4 /* TODO */;
conn_param.retry_count = SDP_RETRY_COUNT;
SDP_DUMP_PACKET(sk, "TX", NULL, &hah.bsdh);
rc = rdma_accept(id, &conn_param);
if (rc) {
ssk->id = NULL;
id->qp = NULL;
id->context = NULL;
}
break;
case RDMA_CM_EVENT_CONNECT_RESPONSE:
sdp_dbg(sk, "RDMA_CM_EVENT_CONNECT_RESPONSE\n");
rc = sdp_response_handler(sk, id, event);
if (rc) {
sdp_dbg(sk, "Destroying qp\n");
rdma_reject(id, NULL, 0);
} else
rc = rdma_accept(id, NULL);
break;
case RDMA_CM_EVENT_CONNECT_ERROR:
sdp_dbg(sk, "RDMA_CM_EVENT_CONNECT_ERROR\n");
rc = -ETIMEDOUT;
break;
case RDMA_CM_EVENT_UNREACHABLE:
sdp_dbg(sk, "RDMA_CM_EVENT_UNREACHABLE\n");
rc = -ENETUNREACH;
break;
case RDMA_CM_EVENT_REJECTED:
sdp_dbg(sk, "RDMA_CM_EVENT_REJECTED\n");
rc = -ECONNREFUSED;
break;
case RDMA_CM_EVENT_ESTABLISHED:
sdp_dbg(sk, "RDMA_CM_EVENT_ESTABLISHED\n");
sdp_sk(sk)->laddr =
((struct sockaddr_in *)&id->route.addr.src_addr)->sin_addr.s_addr;
rc = sdp_connected_handler(sk, event);
break;
case RDMA_CM_EVENT_DISCONNECTED: /* This means DREQ/DREP received */
sdp_dbg(sk, "RDMA_CM_EVENT_DISCONNECTED\n");
SDP_WLOCK(ssk);
if (ssk->state == TCPS_LAST_ACK) {
sdp_cancel_dreq_wait_timeout(ssk);
sdp_dbg(sk, "%s: waiting for Infiniband tear down\n",
__func__);
}
ssk->qp_active = 0;
SDP_WUNLOCK(ssk);
rdma_disconnect(id);
SDP_WLOCK(ssk);
if (ssk->state != TCPS_TIME_WAIT) {
if (ssk->state == TCPS_CLOSE_WAIT) {
sdp_dbg(sk, "IB teardown while in "
"TCPS_CLOSE_WAIT taking reference to "
"let close() finish the work\n");
}
rc = sdp_disconnected_handler(sk);
if (rc)
rc = -EPIPE;
}
SDP_WUNLOCK(ssk);
break;
case RDMA_CM_EVENT_TIMEWAIT_EXIT:
sdp_dbg(sk, "RDMA_CM_EVENT_TIMEWAIT_EXIT\n");
SDP_WLOCK(ssk);
rc = sdp_disconnected_handler(sk);
SDP_WUNLOCK(ssk);
break;
case RDMA_CM_EVENT_DEVICE_REMOVAL:
sdp_dbg(sk, "RDMA_CM_EVENT_DEVICE_REMOVAL\n");
rc = -ENETRESET;
break;
default:
printk(KERN_ERR "SDP: Unexpected CMA event: %d\n",
event->event);
rc = -ECONNABORTED;
break;
}
sdp_dbg(sk, "event %d done. status %d\n", event->event, rc);
if (rc) {
SDP_WLOCK(ssk);
if (ssk->id == id) {
ssk->id = NULL;
id->qp = NULL;
id->context = NULL;
if (sdp_notify(ssk, -rc))
SDP_WUNLOCK(ssk);
} else
SDP_WUNLOCK(ssk);
}
return rc;
}